Analysing the similarity of students’ programming assignments

Abstract

Having many assignments teachers tend to use automated grading systems to make the grading process more efficient. One of the problems while using such systems is detecting plagiarism. Many tools for detecting plagiarism have been developed and are still being improved. These tools can detect the percentage of program similarity, but cannot determine if the similarity is a product of plagiarism or not. In this paper, it is shown that high similarities between implementations are not always a product of plagiarism. Functions which may have a high similarity percentage among students are those for which solution was provided during the course, coding style used during the course, and small functions with a small number of possible solutions. For these functions, it is not possible to determine whether the code was plagiarized only by detecting code similarity.